Uplift modelling is a strategy for predictive Analytics that directly assesses the incremental impact of a treatment (such as a marketing campaign) on a customer’s behavior. This technique has received increasing attention in customer relationship management, especially in churn prediction tasks. Instead of targeting those customers that are more likely to attrite, which is the case in traditional churn prediction, uplift modelling focuses on those customers that would churn and would be retained when targeted with a retention campaign.
Uplift modelling has been successfully applied in a wide variety of domains, including political elections and personalized medicine. However, it has not been used for predicting student dropout, to the best of our knowledge. Student retention is a very relevant topic for Higher Education institutions since they are constantly facing new challenges to meet a growing set of demands as well as political, financial, and social pressures to deliver quality education to students.
There are several triggers that may lead to students dropping studies. These may be individual issues or a mix of problems. Arguably the most relevant one is poor student performance due to lack of motivation and/or the adequate preparation in high school for absolving a given bachelor program.
In this study, we propose an uplift modelling strategy for avoiding student dropout via courses designed to improve student performance, acting as retention campaigns for student dropout. Data from three different Bachelor programs from a Chilean University were collected, in which these courses are offered to all the students. Our idea is to improve the design of these courses by tailoring them to the students that are more likely to be retained with them.
Original languageEnglish
Publication statusUnpublished - 2019
EventFourth Conference on Business Analytics in Finance and Industry - Santiago de Chile, Santiago de Chile, Chile
Duration: 6 Jan 20208 Jan 2020
Conference number: 4


ConferenceFourth Conference on Business Analytics in Finance and Industry
Abbreviated titleBAFI 2020
CitySantiago de Chile
Internet address

    Research areas

  • Uplift Modelling,Student Dropout,Churn Prediction, Analytics

ID: 47575539